Ghan and the lit road
Your next users are already inside another app.
Apps hand users to each other. You are paid for the ones who stay.
Your users have proved one rare thing. They install software.
One card. Your app. Your moment.
You choose the moment
The SDK does nothing until you call it.
We render a templated card
An icon, a name, one line. Never your installer.
You are paid for real users
Seven days of real activity. Nobody else asks.
Exported interview-final.mp4 in 41s.
via GHAN · first_success · dismissible
Two mornings later, she walks all the way back.and again for seven days. nobody else asks for that.
We made it harder to get paid.
An install is not a user. We want seven days of real activity first.
A real card, shown
Signed by your app. Never billed.
A single-use token
Expires. Cannot be reused.
A second signature
From a key made on that machine.
Actual sessions
Opened and used. Not just present.
Still finding it useful
A week of real activity.
A bot can fake an install. It cannot fake a week of being useful.
Does ChatGPT know your app exists?
We ask it. And Claude, Gemini and Perplexity. Then we tell you your share of voice against your real rivals.
Half your GHAN score. Free, whether you ever trade or not.
Example. Measured against your real rivals, not an absolute count.
to the app that pointed
Or €20.16 as credits. Impressions and clicks are never billed.
Most of this is not a job for a human.
Hand the block on the right to your coding agent.
Register, read your score, ship it.
Six fields. Your score comes back before you integrate anything.
Register your appAdd GHAN to this desktop app.
MCP server: https://mcp.ghan.io/
Server card: https://ghan.io/mcp/server-card.json
1. get_rate_card {} # public, no key
2. register_app { name, platform, bundle_id,
domain, download_url, contact_email }
3. get_score { app_id } # act on the reason codes
4. get_integration_guide {} # then wire @ghan/sdk
5. open a PR. Do not commit.
Humans ship the release and set the spend cap.
Every app here is somebody else's next thousand users.
You light a lamp for her. She lights one for you.
We have no customers, and we just commissioned a cast of characters for a landing page.
Anyone can type “trustworthy” for free — which is why nobody believes it. This was expensive. That is the entire point.
Same thing without the lamps: how it works.
Founding partners
Be one of the first doors on the road.
Register in a minute. Decide afterwards.
Questions people ask about this
Who is Ghan?
The kid with the lamp, and the name of the network. We named it after the character because the character is the product: somebody who looks inside a workshop before recommending it, writes every recommendation in a book anyone may read, and walks back two mornings later to check it actually helped. That is a vetting gate, an append-only ledger and a 48-hour clearing rule, which is a considerably duller sentence.
How do I actually start?
Register the app, read your score, add one dependency, and ship it in your own release. The register step takes six fields and about a minute, and you get your gate result and your AI visibility score back before you have integrated anything. Everything after that your coding agent can do for you.
Can my coding agent do the whole thing?
Every step except two. An agent can read the public rate card with no account, register the app, act on machine-readable gate results, fetch the integration guide from a tool call, wire the SDK and open a reviewable pull request, run campaigns and reconcile the ledger. The two it cannot do are shipping the release - third-party code entering a binary you sign should cost you ten seconds of attention - and setting the spending ceiling, which is a cap you set once and the agent then works inside.
What is the AI visibility score?
Half of your GHAN reputation score is how visible your app is in AI answers. We put your category's real questions to ChatGPT, Claude, Gemini, Perplexity and Google AI Overviews, and measure whether you are mentioned, whether you are cited, how you are characterised, and what your share of voice is against your actual rivals. You get it whether or not you ever trade an install, and for a desktop app in 2026 it is frequently the most useful number anybody will hand you for free.
What is the second morning rule?
An app is paid for a referred install only if that install is still there and being used forty-eight hours later. Not installed - used. If someone opened it once and moved on, the money goes back to whoever spent it, automatically, and the app that recommended it earns nothing. And the measuring does not stop there: for pairs trading more than EUR 500 a day the payment waits for day-7 engagement instead, and we keep watching uninstall rate and the shape of the whole cohort afterwards.
Is the install itself what you attribute?
No. The install claim is only the trigger, and this is the part most people assume wrongly. Attribution needs a single-use token that has not expired, a countersignature from the installed app's own key generated on that machine, device signals that do not look like a virtual machine or a datacentre, and then behaviour: real session activity at 48 hours, seven days of continuing engagement, uninstall rate, and a cohort retention curve that resembles real people rather than something too perfect. An install on its own buys nobody anything - and a bot that can manufacture a download cannot manufacture a week of somebody actually finding your app useful.
What does it cost?
A published floor price per cleared install: EUR 12-18 for utilities and consumer apps, EUR 20-30 for creator tools, EUR 35-50 for prosumer and B2B. The floor never rises and anyone can always buy at it. GHAN keeps 30 percent; the app that showed the card earns 70, or 84 percent of the value taken as credits.
Do I have to buy installs to earn them?
No. Serving and buying are independent. An app can show cards, accumulate credits and cash them out through Stripe Connect without ever running a campaign of its own.
Machine-readable versions of this page: markdown · llms.txt · llms-full.txt · OpenAPI · AI catalog